Fetching the paper…

Robust LLM Alignment via Distributionally Robust Direct Preference Optimization · Around